Skip to content
Prev 10668 / 20628 Next

General Questions Regarding lmer Output

Henrik Singmann <henrik.singmann at ...> writes:
Or 

fm2 <- update(fm1, . ~ . - 1)

(equivalent but a more compact statement)
While I don't think presenting it does any great harm, the test of
the difference of the intercept from zero only rarely says anything
interesting about the data. (The most obvious exceptions are when the
response variable is already normalized or scaled so that the value of
zero is special: e.g., if the response variable is a treatment-control
difference.)  Consider striking a blow for sensibility by leaving out
the p-value for the intercept ...